Table 1.

Reducing the frequency of dining out in April-May and media usage: results on Logistic regression

	(1)	(2)	(3)	(4)	(5)	(6)
Daily LINE user	2.03***
	(1.73, 2.39)
	[<.001]
Daily Twitter user		0.86
		(0.72, 1.03)
		[.107]
Daily web aggregator user			1.56***
			(1.28, 1.90)
			[<.001]
Daily TV user				1.52***
				(1.28, 1.80)
				[<.001]
Daily online news user					1.39***
					(1.18, 1.63)
					[<.001]
Daily newspaper user						0.97
						(0.75, 1.24)
						[.786]
University	1.25*	1.25*	1.26*	1.23*	1.22*	1.25*
	(1.02, 1.54)	(1.02, 1.54)	(1.02, 1.55)	(1.01, 1.51)	(1.00, 1.49)	(1.02, 1.54)
	[.035]	[.030]	[.029]	[.044]	[.048]	[.030]
Female	1.39**	1.53***	1.56***	1.48***	1.60***	1.53***
	(1.12, 1.74)	(1.24, 1.89)	(1.25, 1.95)	(1.19, 1.86)	(1.29, 1.98)	(1.24, 1.88)
	[.004]	[<.001]	[<.001]	[.001]	[<.001]	[<.001]
Age	0.99*	0.97***	0.97***	0.97***	0.97***	0.97***
	(0.97, 1.00)	(0.96, 0.99)	(0.96, 0.99)	(0.96, 0.99)	(0.96, 0.99)	(0.96, 0.99)
	[.045]	[<.001]	[<.001]	[<.001]	[<.001]	[<.001]
Prefecture FE	YES	YES	YES	YES	YES	YES
n	2,167	2,167	2,167	2,167	2,167	2,167
Hosmer-Lemeshow p-value	.406	.551	.052	.419	.828	.582

The dependent variable is the answer to the following question: ‘has your frequency of going out for dinners increased or decreased since last March?.’ All the specifications control for prefecture (region) fixed effects. The odds ratios are reported. Standard errors are clustered at the prefecture level. 95% CI are in parentheses. p-values are in brackets. *** p < 0.001, ** p < 0.01, * p < 0.05.